We investigate the importance of accounting for uncertainty a priori in production scheduling in the presence of feedback. First, we examine different optimization models (deterministic, robust, and stochastic programming), used to generate the open-loop schedules and describe the modeling of uncertainty in each case. Second, we present a formal procedure for carrying out closed-loop simulations in order to study and compare the closed-loop performance across the models as attributes such as the demand uncertainty observation horizon, order size max-mean relative difference, and load on the process network are varied. Finally, we analyze the results of the simulations to draw insights on how the above attributes affect the closed-loop performance of the different models across networks and expound on the paradoxes observed.
All Science Journal Classification (ASJC) codes
- General Chemical Engineering
- Computer Science Applications
- Mixed-integer programming
- Online scheduling
- Real-time optimization